- LJ::delete_talkitem() ... needs light delete & strong delete (mark as delete, vs. do it). (accidental always-strong deletions were causing children not to load on talkread pages)
- editjournal_do -- sent back to wrong talkread URLs
- editevents couldn't change metadata (currents*, backdate, etc)
- month subject view: wrong talkread links
- server error when borking talkread URL (adding journal= param to non-clustered case, etc)
- can make lj_ usernames using lj-
- replying to comment from email goes to wrong page
- userpics not loading. (was a problem when in memory cache mode)
- webley style is inherently broken (uses only %%messagecount%% in LASTN_TALK_READLINK for %%readlink%% and tries to make its own talkread URL from %%itemid%%, but now it can't make a correct URL) this would require S2. Disable this style? Update: I know how to fix this now.
- poll images not loading (just not there)
- moods themes/images not there (site-specific & just not there... clustering didn't affect that)
- month view: no subject link when has <lj user=''> in subject? does this also occur on lj.com?
- can't post bogus ljuser tags?
- can't make communities? need more info. clustering should've affect this.
To claim a fixed/unresolved bug above, mail firstname.lastname@example.org .... include a talkread URL proving you were first. He'll feed those to me in a few days.
Now, everybody go test more .... everything above should be fixed. If I forgot anything from the last post, announce it again (after you test it again, though... something above might've fixed it.). Also, when replying here, please stay on topic. I'm getting all comments emailed to me and I have to read through them to find the good stuff.
Update:BTW, testing will go on another several days.... the next interesting stage will be when we convert everybody's cluster 0 accounts to cluster 1. So start making some bookmarks/copies/screenshots of data from your cluster 0 account. No links should break and no data should change.
Another Update: You can go here to meet people: